Regioselective and stereoselective methods for the synthesis of the pentitols

Abstract

Several different approaches to the stereoselective synthesis of xylitol (1), as well as the other two pentitols, ribitol (2) and DL-arabinitol DL-(3), from the (Z)- and (E)-1-hydroxypentadienes (4) and (5) and the (Z)- and (E)-4,5-epoxypent-2-enals (6) and (7) are described. They rely upon either (a) epoxidations of allylic C=C double bonds followed by stereospecific (anti) and sometimes regioselective epoxide cleavages, or (b) syn-hydroxylations of allylic C=C double bonds. Employing approach (a), the (Z)-isomers (4) and (6) do not afford any ribitol (2) among the products and the (E)-isomers do not afford any xylitol (1). The consequences are reversed when approach (b) is adopted. The most convenient synthesis of xylitol (1) starts from the (Z)-isomer (6) of 4,5-epoxypent-2-enal. The formyl group in (6) is reduced, provided acidic work-up conditions are employed, to yield (Z) - (4RS) -4,5-epoxy-1-hydroxypent2- ene (9), which is characterised as its acetate (10). Opening of the epoxide ring in (10) with acetate ion gives the triacetate (11), which is deacetylated to afford a key intermediate, (Z)-(4RS)-1,4,5-trihydroxypent-2-ene (12). Epoxidation of (12) with peracids (e.g. p-nitroperbenzoic acid) yields (t-butyl hydroperoxide with catalytically active Ti4+, V5+, and Mo6+ complexes fails) two epoxides (13) and (14), arbitrarily named isomers A (13) and B (14) subsequently shown to have the relative stereochemistries (2S,3R,4R) and (2R,3R,4R), respectively. Epoxide ring opening with acetate ion in acetic anhydride of the more abundant isomer B (14), obtained with 70% diastereoselectivity, yields xylitol penta-acetate (16) as the major product (>80% diastereoselectivity) along with small and trace amounts of the other two pentitol penta-acetates. Epoxide ring opening of isomer A with acetate ion in acetic anhydride is not a straightforward reaction for the most part and has been found to involve the intermediacy of an isolatable bicyclic orthoester (23) en route to some of the xylitol penta-acetate (16) formed as the principal stable product during this reaction. These variations of approach (a) constitute stereoselective syntheses of xylitol (1), which are claimed to be acceptable on a laboratory scale. They provide a slightly better route than an alternative one involving the transformations (4) → (33) → (34) → (39) → (16) → (1), starting from (Z)-1-hydroxypenta-2,4-diene (4), principally because this particular precursor is less readily accessible than (Z)-4,5-epoxypent-2-enal (6). By contrast, the (E)-isomer (5) of 1-hydroxypenta-2,5-diene is obtainable in high yield from the reduction of vinyl acrylic acid and the analogous transformations [(5) → (26) → (27) → (28) → DL-(5) → DL-(3)] provide a highly stereoselective (91%) synthetic route to DL-arabinitiol DL-(3). Osmium-catalysed syn-hydroxylation of (E)-(4RS)-triacetoxypent-2-ene (22), prepared from (E)-4,5-epoxypent-2-enal (7) in two steps [(7) → (20) → (22)], provides yet another approach to DL-arabinitol DL-(3), but the stereoselectivity (76%) of this oxidation is not as good as that observed for the epoxidation of rel-(3R,4R)-3,4,5- triacetoxypent-1-ene (27) in the above transformation. The synthesis of ribitol (2) by osmium-catalysed synhydroxylation of the (Z)-isomer (11) of (22) was achieved with a modest stereoselectivity of 66% for the oxidation step.